Gas Chromatograph detector calibration

5.1 Experimental procedure for the static analytical apparatus

5.1.2 Gas Chromatograph detector calibration

The compositional analyses of the respective phase samples in this study were performed using a Shimadzu Gas Chromatograph model GC 2010 which was fitted with a thermal conductivity detector (TCD), a three-metre Porapak column and utilised helium as the carrier gas. The response of the detector was calibrated prior to performing any measurements as the instrument lacked the ability to quantify any substance passing through the column without the aid of necessary correlations. The operating conditions of the GC per binary system are tabulated in Table 5-1.

Experimental procedures

41 | P a g e Table 5-1: GC operating conditions for the Shimadzu gas chromatograph (GC-2010).

Binary systems Injector Temp/ K

Oven Temp/ K

Detector Temp/ K

He flow ml/min

Detector current/ mA

𝐂𝐎_𝟐− 𝐂_𝟔𝐇_𝟏𝟒 423.15 343.15 423.15 20 70

𝐂𝐎_𝟐− 𝐂_𝟖𝐅_𝟏𝟖 523.15 498.15 523.15 20 70

𝐂𝐎_𝟐− 𝐂_𝟓𝐇_𝟒𝐅_𝟖𝐎 523.15 498.15 523.15 20 70

𝐂𝐎_𝟐− 𝐂_𝟕𝐅_𝟏𝟔 523.15 473.15 523.15 20 60 𝐂𝐎_𝟐− 𝐂_𝟗𝐅_𝟐𝟎 523.15 498.15 523.15 20 60

Calibration of pure components

The direct injection technique was employed in the calibration process, and as such, syringes were utilised to inject pure components of the particular chemicals used in the study. Hamilton syringes from Sigma-Aldrich of maximum volumes of 500 µl, 250 µl, and 50 µl were utilised for the gases while the liquid component was injected into the GC via SGE syringes with maximum volumes of 5 µl, 1 µl, and 0.5 µl. The accuracy of the experimentalist’s technique in injecting consistent volumes was enhanced by the incorporation of Chaney adaptors (Hamilton syringes) into the gas syringes and repeatability adaptors for the liquid syringes, which allowed a specified volume to be set on the syringe before the loading of the syringe.

The injection of both the gas and liquid components was done at ambient conditions. The gas was sampled directly from the cylinder through a low-pressure regulator which was fitted with a septum on its output port. The temperature at the gas sampling point was measured via a thermometer while the atmospheric pressure was measured by the internal barometer of a Mensor CPC 3000 pneumatic high-speed pressure controller. This then allowed the moles of the gas injected to be calculated through the use of the ideal gas equation. This was a viable assumption since (compressibility (Z) ~ 1) for light gases and the injection was carried out at moderate temperature and low pressure (Williams-Wynn, 2016). The moles of the injected liquid were calculated from the molar density of the liquid, measured at the exact temperature of the liquid through a DSA 5000 density meter. All glass thermometers used had their temperature readings corrected relative to the CTH 6500 WIKA temperature probe.

Experimental procedures

42 | P a g e Upon injection via the injector port of the GC, both the liquid and gas components would elicit a response from the TCD in the form of peaks in a chromatogram. The areas generated by the integration of these peaks were then used in conjunction with the actual number of moles of the injected components to construct first order polynomial curves. Only the gradient from the calibration equation was utilised in the calculation of the number of moles because the value of the intercept was distorted by the dead volume which was noted to be inherent in the syringes, particularly the gas syringes.

The linearity of the calibration was confirmed during the experimental work by altering the ROLSI™ sampling time, viz., increasing the size of the peak areas obtained for the same data set.

The peak areas, however, had to remain within the calibration range for the test to be conclusive.

If the mole fractions remained within the uncertainty, regardless of the different peak areas, the calibration data would be deemed linear.

Calibration of mixtures

During sampling of the equilibrium mixture especially at the low pressure, it is possible to obtain peak areas for the liquid and gas components which are below the generated calibration range. The increase of the ROLSI™ opening time to obtain larger samples might be an option in theory for combating this discrepancy. However, such an act usually results in the distortion of the equilibrium conditions within the mixture and is also prone to produce large GC peaks with tailing which would differ from the sharp GC peaks obtained during calibration, thus rendering the action moot. In this work, the calibration polynomials were extrapolated to peak areas which were well below the calibration range, particularly for the liquid components in the vapour phase. This action was validated by carrying out additional calibrations which utilised mixtures of the pure components with compatible substances, in a bid to drastically reduce the volume of the required substance upon injection into the GC. The available syringes (gas/liquid) were not capable of injecting the necessary minute volumes when pure components were used. Thus the preparation of mixtures was the only viable solution. This then allowed the experimenter to inject volumes of the required components which were below the working range of the syringes, thus facilitating the generation of calibration polynomials in the low peak area regions.

The polynomials generated from the direct injection of the pure components and the mixtures were then compared, and the discrepancies noted to be less than 5%, and thus extrapolation of the pure

Experimental procedures

43 | P a g e component calibration was deemed acceptable. The uncertainty inherent in the mixture preparations was dependent on the Ohaus mass balance which was used to measure the masses of each component and the mole fractions of the aforementioned components. A concise description of the calculation procedure for all uncertainties is published in Appendix C.

For the calibration of mixtures to be carried out effectively, it required the following (a) identification of chemicals which were miscible with the pure components, (b) a resulting mixture with reasonable GC retention time and (c) a mixture which required acceptable GC conditions for separation. Separation of the peaks in the chromatogram was further facilitated by the disparity in the boiling temperatures of the mixture constituents. For the gaseous component, which was carbon dioxide, in this case, helium was the obvious choice for gas-mixture preparation, since it was the GC carrier gas and such produced no traces when it eluted the TCD.

The selection of a suitable liquid component for the fluoro-compound mixtures was not as straight forward. Thus it required the use of solubility parameters to determine which chemical species were miscible with the respective fluoro-compounds. The correlation developed by Hansen (1967) for the solubility of chemical species in solution asserted to the need for a small difference between the solubility parameter values of the two components in the mixture in order to obtain high solubility (x₁) of the solute within the mixture as highlighted in the following equation:

x₁∝ exp [−v₁(δ₁^′ − δ₂^′)ф₂²

RT ^] (5-1)

ф₂ – Volume fraction of the solvent v₁- Molar volume

δ- Solubility parameter

Thus chemical species with Hilderbrand solubility parameter values which were similar to those of the fluoro-compounds in question were selected. Table 5-2 lists the proposed components of each liquid mixture in conjunction with the respective Hilderbrand solubility parameter ranges for the chemical species.

Experimental procedures

44 | P a g e Table 5-2: Constituents of the mixtures and Hilderbrand solubility parameter ranges for the respective

chemical species.

Mixture Fluorinated compound Mixing compound

1 𝐂_𝟓𝐇_𝟒𝐅_𝟖𝐎 m-xylene (𝐂_𝟖𝐇_𝟏𝟎)

a(14.5 – 24.1) MPa^{1⁄2 b}(16 – 16.8) MPa^1⁄2

2 𝐂_𝟕𝐅_𝟏𝟔 Pentane (𝐂_𝟓𝐅_𝟏𝟐)

c(12.1- 12.9) MPa^{1⁄2 d}(12.7 – 16.) MPa^1⁄2

3 𝐂_𝟗𝐅_𝟐𝟎 Hexane (𝐂_𝟔𝐅_𝟏𝟒)

c(12.1- 12.9)MPa^{1⁄2 d}(12.7 – 16.) MPa^1⁄2

a,b,c,dHilderbrand solubility parameter ranges for ethers, cyclic hydrocarbons, fluorinated hydrocarbons, and acyclic hydrocarbons respectively. Adapted from Gwinner et al. (2006).

The carbon dioxide + helium mixture was prepared in a one-litre cylinder which was fitted with an internal magnetic stirrer bar, and the mass of each constituent was determined through weighing the cylinder after loading each component. The resulting mixture was then agitated by an external horseshoe magnet which was coupled to a small D.C motor for mixture homogeneity. The gas mixture was also sampled directly from the cylinder through a low-pressure regulator which was fitted with a septum on its output port.

The liquid mixtures were prepared in small glass vials (8 ml), with the respective amounts of each component determined volumetrically. The number of moles of both the liquid and gas components was determined in the same manner as was previously explained for the pure components.

5.1.3 Equilibrium measurements